Akamai Technologies Inc. Analyst Meeting - Final.(Broadcast transcript)

NOELLE FARIS, SENIOR MANAGER, IR, AKAMAI TECHNOLOGIES, INC.: We have a great program for you guys today. I'm really eager to kind of get it started. For those of you who I didn't get a chance to greet on the way in, I'm Noelle Faris. I'm the Senior Manager, Investor Relations for Akamai.

And unfortunately, we're going to have to jump through Safe Harbor language. So please, bear with me. I'll try to spare you the pain and not read it verbatim, but to hit the high points. These presentation and webcasts contain forward-looking statements as defined by the Private Securities Litigation Reform Act.

You should refer to the risk factors in our 10-Q that describe things that may cause our estimates to be incorrect. All facts and figures in today's presentation are as of November 20th, 2008 and we disclaim any obligation to update these statements going forward. Finally, we will be using some non-GAAP estimates today. You can find the GAAP reconciliations for these measures on the Investor section of our website.

So, as many of you are aware, this is our 10th year and we're excited to be celebrating that. What you may not know is some of the history that has gotten us to this point. And so, we're going to kick off today with a great video. And just sit back, relax and enjoy.

(VIDEO PLAYING)

PAUL SAGAN, PRESIDENT AND CEO, AKAMAI TECHNOLOGIES, INC.: Good morning, I'm Paul Sagan, the President and CEO of Akamai. It's great to see so many familiar faces here on our 10th anniversary to hear our story. I'll acknowledge, there is -- it probably seems a little in congress. But we'll just live in a stock market free bubble for a few hours and talk about our business.

We've been at this for 10 years. We're great believers in our business. We bring great value, to thousands of the leading businesses online. And we'd like to talk to you a little bit about what we do, how we got here and where we're going. You're going to hear about unique architecture, unique services. And from some of our customers, you'll hear about the unique value that we bring to them. And we're very excited about it.

Let me walk through the agenda so you have a good idea of what's coming this morning before lunch. Tom Leighton, our co-founder and Chief Scientist is going to talk to you about Akamai's EdgePlatform and the advantages that we see and that we bring our customers in the unique architecture and design of our system.

Then Mike Afergan, our Chief Technology Officer, is going to talk about how the services that we operate and sell, bring unique value to enable our customers' business models across media and software, commerce and now the new advertising ecosystem and the services that which we've just started announcing in the last couple of weeks. And a new space, generally a new space for us that we're extremely excited about.

Then, we'll have a first Q&A session where Tom, Mike and I will take your questions about the technology and the products. Then Bob Hughes, our Executive Vice President, will talk about how we go to market, the verticals and the products and the solutions and really how we sell. And J.D. Sherman will bring you, probably the part you're most interested in, which is the financial review. And then, we'll have an operational Q&A.

We're also going to have two customer segments today, which I think you're going to find extremely interesting. They're usually the highlights of our customer forums and our investor days. You're going to hear from IAC and Rick Stalzer, who's going to come talk to you about why they're using, not just our traditional content delivery solutions and they've used those for a long time, but our new Advertising Decision Solutions and how they see us helping them make more money with their web properties. So, they'll talk to you about our latest group of offers.

And then later, you'll hear from one of our customers in the eCommerce space, CSN Stores, to talk about how our dynamic capabilities help them sell more online. CSN Stores is one of the largest web-only home good sellers. And they have a very unique set of products and a very interesting business and using our advanced capabilities helps them sell more online. And so, I think you might find that among the more interesting parts of today, and a real highlight.

So, we think about the Akamai difference. We're focused on helping our customers' business models. How do they make more money or take costs out of their business? We've built a reputation over 10 years as the highest quality provider. We are looking at literally billions of dollars of transactions going across our system. The old standard of just good enough or oh, it's the internet, when we began 10 years ago, does not work today.

And so, we have worked really hard. And where our customers make money, they want mission critical performance and that's what we deliver. And it's helped us build trusted relationships with the who's who of internet companies on a differentiated platform that you'll hear a lot about from Tom and Mike.

We help our customers sell more goods and services. One of the reasons they rely on us is they can measure that they sell more. The shopping cart is larger. The likelihood that someone will buy is greater when they use our services. If they're trying to inform or entertain a large audience, they use us for the highest quality delivery where they make money.

If they're trying to market or advertise and their brand is critical, they're using Akamai. And now through our new Advertising Decision Solutions, they have a new way to effectively raise their CPM, to raise the value of their marketing dollars. And we know that online advertising in the last 15 or 16, 17 years has grown into a $20 billion market in the US. But largely, it has not lived up to the great hope of the internet of being targeted and accountable. We believe that our new services are going to provide this value in a unique way and you'll get to hear about that.

You know us for working with people who distribute software in a more traditional way online. Because of our advanced dynamic capabilities, we're now enabling the software as a service market to truly scale and change the way software is sold. And fundamentally, we're about changing the way people conduct business online. Driving revenue up and costs out through the use of the network.

As a company, it's been a pretty interesting year for us. As you know, we've had our ups and downs in a variety of ways and some unimaginable ones. But we've had, for the last number of years, steady growth. And we're very optimistic about our future. I won't spend a lot of time on the numbers here because J.D. will get into it. And certainly, you've -- you are students of all of our filings.

We fundamentally -- we accelerate the delivery of rich media, dynamic transactions and web and IP-based applications. We're now enabling more relevant online advertising for a who's who of customers who access our platform and rely on us to drive their business.

Now, I want to make a comment about some of the restructuring measures that we announced yesterday. I think we have a history of being very prudent managers, very disciplined investors, and people who try not to get too far ahead of our headlights.

We're all aware of what's going on in the world around us. We're in the middle of a global recession. I don't think there's any denying that today. Visibility in the future is as low for, I think, every business as it's ever been. It's fundamentally different than the last time we went through this when the tech bubble burst because the consumer market kept going and most of the other industries were unaffected. And so, we could launch services, and create visibility and predictability.

I don't think that's true for any industry right now. We have a long term view that the internet will keep growing and it's a great opportunity. We're not going to back away from our industry or our commitment, but we're going to be very cautious about how we spend and invest going forward. We're going to focus our investment dollars on the new and the high growth areas. And we're going to be diligent about taking the costs out of our traditional services. We've always been in a market where unit prices had to decline every year and volumes went up.

We think that that demand from our customers will be as strong as ever. So, we're going to focus on efficiency and the new. And so, even with the growth you've seen and with the profitability and the cash generation that this business has demonstrated, we felt it was prudent to take costs out now to make sure that we could continue investing where it made sense.

So, we'll take charges of about $6 million, that includes both the reduction in forest and some other things were detailed in our filings. J.D. can talk about that a little more. We removed 7% of the employees from the Company, a very painful thing to do. But we did it in an Akamai way. We did it quickly. We did it cleanly. We communicated it very clearly to our employees. We told virtually every employee in a 24-hour period yesterday.

So, we're not a company that says we're going to take 10% of our company over the next few months, we'll get back to you. That just creates internal chaos. So, the entire company understands our strategy, where we're going. Not something we wanted to do, but we did it in a very clear and defined way. So, inside the company, everybody now knows what their role is and there is no uncertainty.

We're going to take the savings that we generated both from the reduction in headcount and a series of other cost cutting measures and invest them in what we believe are the high growth areas of our business going forward. Our Advanced CDN Solutions, the Dynamic Site Solutions that you know about, our Application Performance Services which we introduced several years ago, our new Advertising Solutions and expansion in selected international markets where we think there's the highest likelihood of greater than average growth.

And just to underscore how confident we are about the long term aspects of the business, we think that we will end next year with more employees than we'll end this year. So, we took costs out where we didn't think it was key, and we're going to slowly continue to add resources where we think we're going to get the highest value.

And I also want to underscore that we are not changing our business outlook or our guidance. On the last call, we gave you our guidance, we gave you our outlook. We told you how much visibility we have. Nothing has changed that. I don't want you to think we took this action because we have a fundamentally different view. We just have a very cautious approach to the future and how we're going to spend our resources.

So, let's talk about what Akamai does. We effectively go to market in several key areas. Bob will spend more time talking to you about this. But you can think of these as the four largest opportunities for us, Media and Entertainment online, high tech software delivery, SAS, Commerce B2C, and Commerce B2B. We'll spend some time talking to you about customers and those solutions in detail. Here's just a sample of some of the great brands, so that you can place who buys what from us in your mind.

And we sell solutions into industries. And one of the keys for us is, as a software as a service type model, we're doing this on a shared platform. One of the reasons that we could, for example, bring forward our application performance Services as highly profitable, highly scalable services, is we're doing it on the same platform. So, on this deployed technology, we launched our Digital Asset Solutions. You can think of those as content delivery. How do I deliver large files? How do I make a webpage appear?

On top of that, we added our Dynamic Site Solutions for whole site and then Dynamic Site many years back. Really in the last time we were in a financial crisis. Then about two or three years ago, we added our Application Performance Solutions. And then most recently, we announced our Advertising Decision Solutions. And that's where the recent acquisition of acerno fits in. And for us the vision is, as the internet grows and the opportunity grows, we can continue to innovate more software as a service type solutions on the shared platform.

Now, you may say, if you've been around here a long time, I've seen this slide before. You ripped this off, didn't you? Well, in fact, you're absolutely correct. I took this slide from nine years ago. That is exactly the same vision, we took out on our road show for the IPO in 1999. This is a slide from that road show that I went out on.

Back then, we talked about using a shared platform. Back then, it was doing web content. Streaming hadn't even started, this the beginning of dynamic content. We talked about an application delivery platform. We made that real with EdgeComputing, Application Performance, and then we really talked about changing the way the internet works. And I use the expression air traffic control for the internet. That's the path that we've been on for 10 years. And I have no doubt that that's going to carry us for another 10 and beyond.

So, I think as you go out and you hear people change their story, change their strategy, what you'll hear from us is a very coherent and consistent message. Same fundamental view of how the internet works, same view of the technology, and the ability to innovate economically on top of it.

So in summary, Akamai is the technology and the brand leader for internet performance. Our continuing innovation is going to enable us to help customers with the coming waves that will continue to grow of online advertising, commerce and entertainment. We serve customers across diverse markets. And I think that diversification is going to serve us very well in the next one, two or three years. If you know how long this will last, let me know.

And we do it this time with extremely strong financials. The last time it was a battle for survival. Today as you know, we have a huge amount of cash available to us. We generate a tremendous amount of cash flow and we're very profitable. It gives us the ability to continue innovating where we see opportunities.

So, launching the Advertising Decision Solutions which Mike will talk about was a two-year effort. And you could say, well, why are you going to do it now if we're going to be in this tough period? Online advertising isn't going to go away. It may grow more slowly. But it's a gigantic market and in this environment, we think we can demonstrate unique value that will help us grow there. And our strong balance sheet will let us continue to move in the areas where we think there's the greatest opportunity.

So, thank you for coming, I hope you enjoy the morning. Stay for the lunch if you'd like. We think it's going to be very informative. Love to take your questions at the right time during the day. And it's my pleasure to introduce, for those of you who don't know him, Tom Leighton, co-founder of Akamai, our Chief Scientist and my partner, who's going to talk to you about the EdgePlatform. Tom.

TOM LEIGHTON, CHIEF SCIENTIST, AKAMAI TECHNOLOGIES, INC.: Thanks very much. This morning, I'm going to talk about the two basic approaches used in the market place for content delivery. The traditional big data center approach and the massively distributed approach used by Akamai.

I will talk about the key differences between these approaches and try to explain why we believe that the massively distributed approach affords greater speed, reliability, scale and less cost. I'm going to address head on several of the questions that you've posed to us. And I'll also talk about handling dynamic content, why the massively distributed approach is uniquely well-positioned to do that well.

Akamai's approach to content delivery has been to deploy a distributed platform at the edge of the internet, close to the end users. And when I say distributed, I mean massively distributed. Today, we have 40,000 plus servers. By itself, that's not such a striking number and not so important. What's important is that those servers are in over 1,500 different locations spanning 950 networks in over 650 cities in 69 countries.

This is what I mean by massively distributed. Not 40,000 servers but where the servers and the software on them is located. Our goal is to have our servers be so close to the end users that wherever our customers' users are, and however they connect to the internet, there's an Akamai server across the last mile connection in their city in their network.

Akamai's approach to content delivery is very different than the approach followed by our competitors. They all use what's called the big data center approach. A very traditional approach where you take your CDN servers and put them in a few large data centers at the core of the internet, then you connect those servers with transit pipes from one or more transit providers and then the transit providers get the content delivered to the end users.

Today, there's about 50 companies that use this approach to compete with Akamai. In fact these days, it seems like there's another one being announced every day that somebody's going to put some servers in a data center and open up shop to compete with Akamai. Now, this has given rise to concern about competitive pressure on Akamai. For example, it's not unusual to see quotes like these in the press. "Competition in the content delivery network space is getting hotter every day. The flood of competition is placing Akamai under enormous pressure."

You show us these quotes. We read them all the time. And particularly AT&T, in particular, will enter the content delivery in a streaming media market, unveiling a plan Monday, to provide services for companies to create, manage and distribute audio and video over the internet. AT&T is going to spend $200 million building the infrastructure for the plan, servers, load-balancing equipment and the like.

Now, those of you who have followed us for many years know that competition is nothing new for Akamai. In fact, those of you who were following us back in 2000, you know that we had 50 competitors just like this in 2000. In fact, these quotes are not recent. These quotes are eight or nine-years-old, even the AT&T quote.

Now, you've all read recently AT&T is getting in the business. They're going to compete with Akamai. Well, this time they said $70 million. You know, eight years ago they said they're going to spend $200 million to compete with Akamai. And they do have a CDN, they've had a CDN all along. It's not a major factor against us in the market place.

Now, the names of the competitors, of course, have changed. Almost all the original 50 are gone and they've been replaced now by a new set of 50. But the concerns they raise, and that you are concerned about, are very similar to the issues that were raised back then.

Of course as Paul mentioned, Akamai is in a much stronger position today than it was in 2000. Of course in 2000, we were losing money. We didn't have very many customers. We were just getting started. Today, we're very profitable. We provide our services to a large and important customer base. And we've developed a variety of new solutions that enable us to go well beyond what the 50 competitors can do selling ordinary CDN.

That said, the basic issues are the same. The basic approaches are the same. If you go back to 2000, the 50 competitors then all were using the big data center approach. Now in fact, if you look back then, our most significant competitor in that time frame actually had more deployments, more locations than our competitors today. Today, the -- of the 50 competitors, the one that has the most locations has about 20. They talk about 50 logical locations, but physical locations, they have less than 20 today. In 2000, our most significant competitor had 30 locations. So, it's still the same approach.

Akamai's approach is also the same. There's no change in the basic approach since way back when in 2000. However, we have scaled. At the beginning of 2000, we were in 60 locations. Today, we're in 650 cities and 1,500 locations. The number of locations has increased by a factor of 25 in the Akamai platform, and when we get here together again in five years, that number's going to be bigger still.

Now, many of you ask us why? Why does Akamai put our servers in so many locations? After all, everyone else is using the big data center approach. Why are you doing this? Isn't the other way really the right way to go, the way the other 50 guys are doing?

As I'm going to explain today, the answer is no. By deploying our platform close to end users, at the edge of the internet, we're able to achieve the very best performance, reliability, scale and cost. The reasons are the same as they were 10 years ago when we got started. In fact, they're even more compelling today than they were in 2000.

But let's start by looking at specific questions that you ask us. First, does the Akamai approach, massive distributed location of our platform, does it deliver meaningfully better performance? Or does the big data center approach do just as well? Or even if it doesn't do as well, is it good enough?

Second, is the Akamai approach preferable for the widespread delivery of large media files? We hear a lot that yes, Akamai's good for eCommerce and small objects, but that big DVD, forget it. It won't fit in the Akamai server. One of the craziest things I've ever heard. And you can't deliver a lot of them, that the big data center approach is ideally designed to deliver big media.

Isn't the Akamai approach more costly? After all, you've got servers everywhere. That must be a nightmare to maintain and manage? And finally, this is a little bit down in the weeds a little bit, does the big data center approach maximize origin offload? In other words, if you use the big data center approach, does the customer have to do less of the work? And I'll get into details of what I mean there. Is that a more effective approach for offloading the customer? We hear that question a lot.

All right, let's start with the first question. Does the Akamai approach deliver meaningfully better performance? The answer is yes and there's a very simple reason why. If you look at the big data center approach, the content is served from a few big data centers at the core of the internet that are typically far from the end user getting the content.

In the Akamai approach, we're in over 1,500 locations in over 900 networks. These are the last mile networks. We are very close to the end user, often within a few miles. The goal is to be within 10 miles and never to be over 100 miles away.

Now, when you're closer, that means it's going to be faster. You have less latency. And this is born out by lots and lots of head to head competitions and case studies. This is data that compares Akamai head to head against our seven leading competitors in terms of speed. And it's for the first eight months of 2008 and it's comparing how much slower the competition is over the first eight months of 2008 compared to Akamai.

So, Akamai is here, and for example, the blue guy here started being three times slower, got a little better, then had some bad problems occur in the summer, got up closer to four times as slow, right. So, this is data that we keep track of for the seven largest competitors. It has all the names that you ask us about. We didn't hide any names here. These are the fastest and the biggest, most famous competitors to Akamai.

And that -- 2x is a lot and 3x is a lot. Imagine that a customers' leaving Akamai to go to a competitor. Their website's going to get three times slower. The CEO is going to notice that. They're users are going to notice that, three times slower. That's not acceptable.

And it's not just speed, it's also reliability. Take the same set of competitors and measure error rates, where they can't deliver something for some reason. Something just goes wrong on the internet. And huge factors increase in error rates if you switch to a competitor. Anywhere from 3 to 15 times -- Akamai is 3 to 15 times better than the competition in terms of reliability, big numbers there. You know companies will kill for a factor of two improvement in reliability.

Now, this matters when you're doing transactions. This is a live test that was done for a customer using a competing service. And the blue is the time it took to do a transaction using the origin without Akamai. The orange is the time it took using Akamai. Now, the way this is set up, it's a histogram. This is the number of seconds to do the transaction and the height of the bar is the percentage of the end users that would take that long to do the transaction.

So, you can see that with Akamai, they're able to get that transaction. And this is a four-step transaction done in about 12 seconds. And very few end users took beyond 20 seconds. Well, without Akamai look what's happening. The typical users taking 30 or 32 seconds, three times as long. And many of them, you've got a quarter of the users taking longer. Who knows how long, but unacceptably long?

So, the whole point of Akamai is that we take these very bad transaction times, and move them in here to very good transaction times and that gives better revenue for the site. And I'll give you a lot of case studies later that demonstrates that.

Now, a lot of folks say yes, yes, okay, in eCommerce, we get it. You know, that's why eCommerce companies use Akamai. But in media, where a lot of the traffic is, performance doesn't matter, that's not true. In particular, latency governs throughput when you're doing downloads. The farther away you are, the less throughput you get. And that impacts the time it takes to download a DVD, the classic large media file.

So, this is data collected for the time it would take to download a 4 gigabyte DVD, a typical DVD, two hour movie. If you're serving it locally, 10 to 100 miles and you have a big last mile connection. So, that's not the bottleneck, in this case, we're assuming it's big enough to handle. We can deliver at 44 megabytes a second in the 10 to 100 mile range. And that takes 12 minutes to download the DVD. The competition is sitting out here, depending on how good they are. And the best they can do is 2.2 hours. And right away, that's a problem. Because once you get over an 1.5 hour for a two-hour movie, you are not real time.

In fact, we did a case study with probably one of the world's largest media companies that wanted to download DVDs based in Tokyo. And they did a head to head with Akamai against a leading competitor and because they said there's a price difference and should they be paying Akamai more? And we said yes, because we're going to enable you to download your videos in real time. That your customer can click and start watching right away, DVD quality and not have a problem. The competitor is going to have to wait hours if you use them before you can watch your movie.

And in fact, we did this study. The competitor took eight hours to download the movie. Akamai didn't get 12 minutes because the last mile pipe was smaller, but we had about a half hour delivery time. Because the last mile couldn't accommodate the speed we could deliver.

Now, half an hour is good enough to watch in real time because you start and you're downloading it faster than you're watching it, because you can watch within a second, okay. Where as here, you're going to wait seven hours before you start watching your movie, and that's a qualitative difference. And so, that's a very happy Akamai customer today.

So even in media, it matters -- so, that gets down to the question, okay, when is good enough really good enough? And this is a question that's faced in the market place every day. Well, if it's not performing very well, good enough can be okay if you're not trying to monetize your content, if your brand image is not that important. If you don't care that your site's slow, your downloads take a long time, or if your consumer experience doesn't really matter, that you don't mind that they wait seven hours to get their movie.

Now, in those circumstances, and they do arise, the competition can be good enough. And they go get business. Now, we find often if the customer falls in those three categories, they're also a customer that doesn't want to pay very much, and may not be willing to pay anything at all because it just isn't valuable to them. And so, our customer base is focused on websites that -- for which the brand is important or they're trying to make money, or the consumer experience matters to them. And those are the ones that are willing to pay money for it.

All right, so that's the first question. The Akamai approach does deliver meaningfully better performance and it matters. Now for the next question, is the Akamai approach preferable for widespread delivery of large media files? Or is the big data center approach really optimized for that?

Well in fact, the big data center approach, I would say, is the reverse of optimized for that. A big data center is the last place in the world you want to serve large media files from, especially if you're serving a lot of them, because they're on the wrong side of the bottlenecks at the peering points of the internet.

Like you see this one, there's big news events or a hot video becomes available. If they're being served on the wrong side of the bottleneck, like not at the edge, then you get clogged up here and you get horrible performance to the end users. They just can't get the bits through the internet. In fact, this phenomenon has led a lot of the pundits to say the internet will not scale to handle large, widespread distribution of large media files. That it can't be done for this, because of this very phenomenon.

And it surely cannot be done from the big data centers. And we'll look at this more carefully. Here, you can do a few big files from there and at levels that people are looking at today. You can do it. Once you start getting traffic spikes, it's a problem.

Now, Akamai on the other hand follows a different strategy. We have some servers in big data centers, but most of our servers are in the last mile ISPs, where we can take advantage of the large bandwidth that now exists in the last mile. By being close, we're beyond the bottlenecks. We can fill those pipes here and not get clogged up at the peering point bottlenecks.

Now, if this matters today, it's going to matter even more in the future. Today, it's estimated there's about 400 million people that see video over IP, over the internet. And on average, and this is a very rough statistic, it's guessed that they're watching on average about 10 minutes of video over IP daily. Maybe just a quick hit video or one or a few people watches a TV show or a movie. But on average, 10 minutes of video over IP. And the quality is very low on average, about 300 kilobits a second. That's sort of a thumbnail kind of a thing, a grainy YouTube image, not high quality.

TV quality is -- ordinary TV is about 2 megabits a second. This is a tiny fraction of that. If you put all that together, it's not an insignificant amount of traffic, but it's less than a terabit a second, less than 1,000 gigs of traffic. And that is insignificant compared to where we might be headed. And that we believe ultimately will take place as much more video moves over IP.

Going forward you'd say, well there's billions of people watching videos, say 1 billion actually watch on the internet. And today, we serve 1 billion people every day. Not all video, but we have 1 billion people coming to Akamai for content that we deliver today.

Now, the average person watches a stunning amount of TV today. I mean, the stats -- it's hard to believe, but it's something like four to six hours. Say that just two hours of that one DVD worth at night is done over IP. Now, of course for that to happen, the quality has to be a lot better. DVD quality is about 4 to 5 megabits a second, true HD, true Blu-ray. True Blu-ray is about 20 megabits a second.

Now, say that when you average it all together that the average person who's watching two hours is going to get 6 megabits a second of quality. That's a little better than DVD. People call it HD. It's not really HD, but it's in between. So, you get that average level of quality.

Now, you multiple those numbers together and you produce 500 terabits a second of traffic demand. That's 500 million megabits a second. That's a lot, and it's a big increase over where we are today. And this is what we think about when you think about the explosion of media traffic over IP. And it's why the pundits say the internet's going to collapse. It cannot handle widespread distribution of big media files. That's just -- it's enormous. We're already at the straining point with the peering points being choked.

All right, well now you got to see the two approaches to CDN out there. Can either one can handle this problem? Well, let's look at the big data center approach. And for now, let's ignore the peering point problem which is the first bottleneck that they run into. Just look at the big data centers.

Adding a data center for those companies is expensive and painful. Their systems are not designed to have large numbers of them. But assume they grow to 50 locations, 50 big data centers. Assume each one gets 20 pipes with 10 gigs each. Now, a lot of data centers you're lucky to get one pipe of 10 gigs. But say you got 20 pipes of 10 gigs. What do you got?

Now, you do the math, you find out they can grow to 10 terabits a second doing that. In fact, they can't, in reality, they can't even get that far because the peering points are going to kill them. They can get 10 terabits out of their data centers. They're not going to get them across the internet into the homes or the businesses. But even if the data centers, 10 terabits looks like a problem. Once you get there, that's not going to cut it. That's 2% of the market demand if you get widespread use of media over IP.

All right, what about the massively distributed approach? So here, assume that Akamai grows the number of locations further. And we have plans to do that. And say we go to 10,000 locations and at each location, we're deep now into the networks close to the end users that we have 25 gigabits a second of location -- per location, not that much, maybe a dozen or two servers in each location.

And of course, as you know, we're working on peer to peer to support us. And say we get a factor of two boost from the support for peer to peer for the most popular downloads. Put all that together and you get 500 terabits a second. It is doable. In fact, it is the only way it's going to be done. Whether it's Akamai or somebody else tries to do exactly what we do, it's the only way you can scale the internet to deliver real video over IP. It can't be done from core data centers.

In fact, we're delivering a bunch of large media files today. In fact, we're delivering the only one out …